[
https://issues.apache.org/jira/browse/HADOOP-7269?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13033323#comment-13033323
]
Todd Lipcon commented on HADOOP-7269:
-------------------------------------
This opens up scope creep that makes me a little bit nervous. If we add the
ability to set metadata in create(), then we probably also need the ability to
query metadata, modify metadata on existing files, etc, which is all of a
sudden a lot of code.
There's a trade-off between least-common-denominator APIs (sucks for everyone)
vs support-all-FS-features (sucks for compatibility). I think this might be
going too far towards the latter.
Would it make sense to add these functions directly to the S3 filesystems but
not to the overall API? Then if a user really needs to code specifically
against S3, they can down-cast to S3FS and do what they like.
> S3 Native should allow customizable file meta-data (headers)
> ------------------------------------------------------------
>
> Key: HADOOP-7269
> URL: https://issues.apache.org/jira/browse/HADOOP-7269
> Project: Hadoop Common
> Issue Type: Improvement
> Components: fs/s3
> Reporter: Nicholas Telford
> Assignee: Nicholas Telford
> Priority: Minor
> Attachments: HADOOP-7269-S3-metadata-001.diff,
> HADOOP-7269-S3-metadata-002.diff
>
>
> The S3 Native FileSystem currently writes all files with a set of default
> headers:
> * Content-Type: binary/octet-stream
> * Content-Length: <computed from object size>
> * Content-MD5: <computed from object data>
> This is a good start, however many applications would benefit from the
> ability to customize (for example) the Content-Type and Expires headers for
> the file. Ideally the implementation should be abstract enough to customize
> all of the available S3 headers and provide a facility for other FileSystems
> to specify optional file metadata.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
